Chapter question: The lower half of this patient is bright pink and the upper half is not. How do I know, how urgently must I act, and why does every fix cost me something else?
Evidence search date: 7 September 2026.
Primary sources: ELSO Red Book 6th edition Ch 2 (nomenclature and differential carbon dioxide), Ch 4 (hybrid configurations), Ch 18 (paediatric cardiac), Ch 28 (adult cardiac management and ventilation); ISCCM Manual Ch 25, Ch 30 and Ch 41; Taha Ch 6 and Ch 7; ECPR and Resuscitative ECMO Ch 7; Szuldrzynski 2024.
External evidence retrieved this session: one randomised experimental animal study, the largest published clinical series, a unit guideline updated in August 2026, and a set of case reports describing forms of the syndrome that the textbooks do not.
This chapter contains a direct conflict with Chapter 15. Chapter 15's first medical step for a distending ventricle is to lower ECMO flow. The classical first step for differential hypoxaemia is to raise it. §16.8 sets out the conflict, the evidence on each side, and what to do when one patient has both problems — which is common, because they have the same cause.
What this chapter covers — and what it does not
This chapter owns | Deferred to |
The syndrome itself: why it appears, who gets it and how often; the sampling-site problem and the monitoring set; the isolated cardiac and radiological forms; the differential diagnosis, including differential venous drainage and right ventricular failure; differential carbon dioxide; the management ladder and the flow conflict at its centre; how to choose between reconfiguration options; prevention at cannulation | The mixing point as a moving balance → Chapter 13 §13.2
The reverse harlequin effect → Chapter 13 §13.4
LV distension and unloading → Chapter 15
Arterial cannulation site selection → Chapter 12
Blood gas sampling protocols → Chapter 27
Cerebral monitoring in depth → Chapter 28
Oxygenator failure → Chapter 32
Building and running hybrid circuits → Chapter 78
Ventilation strategy on ECMO in full → Chapter 53 |
16.1 The complication that means your patient is getting better
Differential hypoxaemia — Harlequin syndrome, North–South syndrome, dual circulation, watershed phenomenon — is the one complication in this book that is a marker of recovery. It cannot happen in a heart that does not eject. It appears precisely when the ventricle starts working again, and the better it works, the worse the syndrome gets.
Physiology — the same two facts, arranged the other way round
Chapter 13 §13.2 established that femoral return creates a watershed whose position is a moving balance between native output and circuit flow. Chapter 15 took one consequence of that balance — a ventricle that cannot eject against a pressurised aorta. This chapter takes the other consequence, and it belongs to the opposite patient.
The ISCCM manual locates the watershed anatomically: it "typically occurs between the ascending aorta and renal arteries, but its position depends upon the LV output and the ECMO flow." Everything proximal to it is perfused by the left ventricle; everything distal by the circuit.
The syndrome therefore requires three things at once, and it is the conjunction that makes it uncommon rather than universal:
1. A ventricle that ejects. Enough stroke volume to push the mixing point down the aorta and past the arch vessels.
2. Lungs that do not work. The blood the ventricle ejects has been oxygenated only by the native lung. If that lung is healthy, a distal mixing point is harmless.
3. Femoral — that is, retrograde — arterial return. Upper-body return abolishes the problem by construction (Chapter 12).
Taha adds the modifiers: the picture is "determined by native lung function, volume status, and cardiac function."
Danger — the territory above the watershed is the heart and the brain
The region proximal to the mixing point is not a random half of the body. It is the coronary arteries and the brachiocephalic vessels — the heart you are trying to recover and the brain that determines whether recovery is worth having.
The Red Book states the consequence directly: hypoxaemic blood reaching the proximal aorta and its branches is "potentially causing significant ischemia to heart and brain," and the phenomenon "is more pronounced with poor pulmonary gas exchange and preserved or improving left ventricular ejection upon cardiac recovery."
And it is fast. The one controlled experimental study of this syndrome (§16.8) found histological cerebral injury after only a few hours of differential hypoxaemia in an animal model.
Treat a new right-sided desaturation on femoral VA ECMO as an emergency, not as a gas to repeat in an hour.
16.2 How common is it?
Evidence — the incidence figure everyone quotes, and what it is worth
The number in circulation is 8.8%. It is stated in a 2024 Critical Care monitoring paper — "the incidence has not been clearly stated, but it has been reported to be 8.8%" — and repeated in the 2025 case literature. The Red Book's adult cardiac chapter gives a figure of the same order for peripheral VA ECMO.
Certainty: very low. Apply Chapter 12's rule — two sources agreeing is not two sources. Both modern citations attribute the figure to the same upstream reference, which was not retrieved in this session, and the 2024 paper's own wording concedes that the incidence "has not been clearly stated."
There is a deeper reason to distrust any single number. There is no agreed diagnostic threshold (§16.3); the syndrome is often transient — the largest clinical series found a median time to resolution of 4.5 hours — and it is only detected if somebody is sampling from the right arm. An incidence figure for a condition with no definition, a short natural history and a site-dependent diagnosis is a rate of detection, not a rate of occurrence.
What to take to the bedside: assume it is commoner than 9% in any femoral VA ECMO patient with lung disease, because the ones nobody sampled for were never counted.
16.3 The sampling-site problem
Everything else in this chapter follows from one fact: on femoral VA ECMO, "the arterial blood gas" does not exist. There are several different arterial blood gases in the same patient at the same moment, and which one you get is decided by where the needle goes.
Sampling site | What it actually reports | Use |
Right radial artery | The brachiocephalic trunk — the first branch off the arch. The closest available surrogate for coronary and cerebral oxygenation | The reference site. Chapter 12 places this line at cannulation for exactly this reason |
Left radial artery | The left subclavian — the last arch branch. Intermediate in the mixing gradient | Paired with the right radial, it localises the mixing point relative to the arch branches |
Femoral artery, or any lower-body site | Essentially circuit blood | Tells you the oxygenator is working. Tells you nothing about the brain |
Post-oxygenator circuit sample | Membrane lung performance alone | Chapter 32. Szuldrzynski's pragmatic target is a post-oxygenator PaO₂ of 150–300 mmHg in selected patients, titrated on the blender |
Clinical pearl — three gases at once localise the mixing cloud
The ECPELLA case literature describes the manoeuvre explicitly: simultaneous samples from the right radial artery, the left radial artery and the ECMO arterial line locate the mixing cloud along the aorta — and the authors argue that focusing on the position of that cloud, rather than on the isolated flows of the circuit and the micro-axial pump, makes the decision about how to modify those flows easier.
The logic is arithmetic on three points along one vessel:
— Right low, left low, circuit high → the mixing point lies distal to the left subclavian. The whole arch is native. This is the severe form.
— Right low, left high, circuit high → the mixing point sits between the arch branches. Only the brachiocephalic territory is affected — but that is the brain and the right arm.
— All three high → no differential. Look elsewhere for the hypoxaemia (§16.5).
The right radial number alone tells you the syndrome exists. The three together tell you how much of the aorta you have lost.
Danger — a normal right radial gas does not exclude the syndrome
The mixing point can sit proximal to the brachiocephalic trunk, in the ascending aorta itself. Then the only territory receiving native blood is the coronary circulation, and every arterial sample in the body — right radial included — is reassuring.
A 2025 case report describes exactly this: a man with sepsis-induced cardiogenic shock and ARDS on VA ECMO in whom "the mixing zone of blood from the heart and the VA-ECMO was in the ascending aorta," so "hypoxia was limited to the heart." The presentation was ST-segment elevation, then recurrent ventricular arrhythmia progressing to refractory ventricular fibrillation. A second return cannula was placed in the jugular vein; the fibrillation defibrillated once venous return was added, the ST elevation receded within an hour, and systolic function was normal by day 26.
The authors' warning is the one to carry: isolated cardiac Harlequin syndrome "can be overlooked or misinterpreted as result of coronary artery disease."
This is Chapter 13's pink-patient-with-an-ischaemic-ECG in its extreme form. On femoral VA ECMO, new ST elevation or a new electrical storm is a differential-hypoxaemia diagnosis until the coronary supply is proven — and the ECG may be the only monitor that sees it.
16.4 The monitoring set
No single monitor is sufficient, and the reason is worth understanding rather than memorising: each one fails in a different direction, and on VA ECMO several of them fail at once.
Monitor | What it adds | How it fails |
Right radial arterial line | The reference measurement. The Red Book lists right radial arterial blood as a surrogate marker of coronary oxygenation; ELSO's adult VA guidance is reported as suggesting monitoring right radial PaO₂ to detect differential hypoxaemia | Intermittent. A gas every four hours cannot see a syndrome whose median duration is under five. And it misses the isolated cardiac form (§16.3) |
Pulse oximetry on the right hand or right ear | Continuous, free, and the Red Book names it explicitly alongside forehead NIRS as the way differential hypoxaemia is "easily monitored" | It needs a pulse. Chapter 13 established that pulse pressure falls as circuit flow rises. In the low-pulsatility patient the probe may read nothing, or read badly — and the situations that produce the worst differential are not always the low-pulsatility ones, so this failure is intermittent and unpredictable rather than absolute |
Forehead near-infrared spectroscopy | Continuous, independent of arterial pulsatility, and regional — it measures the territory you actually care about. The Red Book pairs it with right-arm oximetry as the standard detection combination | Taha's monitoring table lists the confounders plainly: NIRS is affected by MAP, PaCO₂, ECMO flow, SpO₂, haematocrit, temperature, sedation and activity, and normal is device-dependent, typically above 60%. It is a trend monitor, not a measurement |
Echocardiography | Answers the question that generates the syndrome — is this ventricle ejecting, and how much? Also excludes the differentials in §16.5 | Intermittent, operator-dependent (Chapter 26) |
The 12-lead ECG | The only monitor that sees isolated coronary Harlequin (§16.3) | Non-specific — and the differential it raises is acute coronary occlusion, which invites the opposite intervention |
Clinical pearl — the reason NIRS earns its place here is that it does not need a pulse
A 2024 Critical Care paper making the case for radial near-infrared spectroscopy as a Harlequin monitor sets the argument out cleanly. Blood gas analysis "is the only gold standard" but is invasive, costly and lagging — it cannot be continuous. Pulse oximetry "depends on the pulse of the fingertip artery, so it is not suitable for hypoperfusion and ECMO advection." Near-infrared spectroscopy, by contrast, is "noninvasive, real-time and continuous... independent of arterial pulse."
That last clause is the whole point, and it generalises beyond the specific proposal: the physiological state that produces differential hypoxaemia is one in which the standard continuous oxygenation monitor is least reliable. A patient on high circuit flow with a narrow pulse pressure has both the greatest risk of upper-body hypoxaemia and the least trustworthy SpO₂.
Certainty: low for radial NIRS specifically — the proposal was retrieved as a correspondence-style paper and its validation data were not retrieved in this session. Certainty: moderate for the underlying point that pulse oximetry degrades with low pulsatility, which follows from how the instrument works.
What to take to the bedside: if the right-hand SpO₂ trace looks poor, do not conclude that the saturation is unmeasurable. Put a NIRS sensor on and sample the right radial line.
Pitfall — the mixing cloud on a CT scan looks like a thrombus
A 2025 case describes an ECPR patient scanned for suspected aortic dissection. The images showed "a sharp demarcation of the mixing cloud with contrast from the descending aorta up the aortic arch," which "raised concerns for a possible thrombus at the aortic arch level." Repeating the study with ECMO flows turned down, allowing contrast through the native circulation, excluded both dissection and thrombus — and, incidentally, demonstrated the impending syndrome.
Two lessons. First, a filling defect at the arch on a femoral VA ECMO patient is a mixing artefact until proven otherwise, and the test is to repeat the acquisition at lower flow. Second, and more useful: contrast-enhanced imaging is an unintentional map of the watershed. If the demarcation sits at the arch, the arch vessels are being perfused natively.
16.5 What else makes the top half hypoxaemic
A low right radial saturation on VA ECMO has a differential, and two of its entries are commonly missed because they mimic the syndrome exactly.
Alternative | How it differs | What to do |
Oxygenator failure | Every site is hypoxaemic, including the post-oxygenator sample and the lower body. There is no gradient | Chapter 32. Check the post-oxygenator gas first — it separates the two diagnoses in one sample |
Differential venous drainage | A distinct entity described in a 2025 ASAIO case series, arising from the circuit's haemodynamic effect on the patient's own venous return rather than from arterial mixing. The authors report that it "can produce symptoms similar to" differential oxygenation and set out how to distinguish them | Recognise that it exists before reconfiguring an arterial limb for what is a venous problem. Certainty: very low — a case series |
Right ventricular dysfunction | A 2021 Perfusion case makes the point in its title — "do not forget the right ventricle." Harlequin syndrome is usually attributed to impaired gas exchange; here it was caused by RV dysfunction and was corrected by conversion to a veno-veno-arterial (VV-A) configuration — that is, by improving drainage, not by adding a return limb | Echocardiography. If the RV is the problem, the third cannula belongs on the drainage side (§16.9) |
Arterial cannula malposition, or an intracardiac shunt | Both appear in the published differential for early severe hypoxaemia on VA ECMO. Echo and imaging separate them | Chapter 12; Chapter 26 |
Reverse harlequin | The circuit-perfused territory is the hypoxaemic one. Caused by interrupting sweep gas on a VA circuit (Chapter 13 §13.4), or by loss of gas supply | Never interrupt sweep gas on a VA circuit. Chapter 13 carries the full warning |
Recirculation in a hybrid circuit | Only relevant once a venous return limb has been added. High circuit inlet saturation with a persistent gradient | Chapter 78. The Alfred guideline advises watching circuit inlet venous saturation, noting that values above 85% may contribute to lower oxygen delivery |
Clinical pearl — one sample splits the differential in half
Before doing anything else, take a post-oxygenator circuit gas and a right radial gas at the same moment.
— Post-oxygenator high, right radial low → the machine is working and the problem is distribution. This is differential hypoxaemia or one of its mimics.
— Both low → the machine is not working. Go to Chapter 32.
It takes two syringes and it prevents the commonest wrong turn in this presentation, which is reconfiguring a circuit whose oxygenator is failing.
16.6 Differential carbon dioxide — the half of this syndrome nobody samples for
The literature is preoccupied with oxygen. The same anatomy produces a carbon dioxide gradient, and the Red Book's nomenclature chapter is unusually candid that it is under-recognised.
Physiology — two lungs, two carbon dioxide tensions, and a control loop between them
The Red Book states it plainly: "Upper body PaCO₂ is determined by lung ventilation and lower body PaCO₂ by circuit sweep gas flow." It calls this differential carbon dioxide tension, notes that it "may easily pass undetected depending on where blood gases are sampled," and adds the variable that makes it genuinely strange — "which regional blood stream supplies the respiratory center in the brainstem."
Read that clause again. The brainstem sits in the upper-body territory. So on femoral VA ECMO with a distal mixing point:
— The respiratory centre is perfused by native, lung-derived blood, and therefore sets the patient's own ventilatory drive from the upper-body carbon dioxide tension.
— That drive changes minute ventilation, which changes upper-body PaCO₂ — a normal loop.
— But the sweep gas independently sets lower-body PaCO₂, and the two are not coupled at all.
The Red Book lists what the balance depends on: "the interaction of the patient's own ventilation regulated by the respiratory center, degree of sedation and neuromuscular blockade, ventilator settings, cardiac output, and ECMO blood flow." Hypocarbia or hypercarbia may develop in one region or the other.
Certainty: low. This is a mechanistic account in a textbook chapter, cross-referenced to material on the ELSO website; no clinical outcome data on differential carbon dioxide were retrieved.
Danger — very low transpulmonary flow makes the natively ejected blood alkalotic
The Red Book's adult cardiac chapter turns the mechanism above into an explicit instruction, and it reuses a rule Chapter 13 already gave for a different purpose:
"Critically low transpulmonary blood flow (<1 L/min), as indicated by arterial pulse pressure <15 mmHg or end-tidal CO₂ <14 mmHg, should prompt adjustments in minute ventilation and PEEP to avoid blood flow stagnation and severe alkalosis of natively ejected blood."
The arithmetic is unforgiving. If only a trickle of blood is crossing the lungs and the ventilator is still delivering a normal minute ventilation to them, that trickle is stripped of carbon dioxide — and it is the blood that reaches the coronaries and the brain. Severe alkalosis constricts cerebral vessels, which is precisely the wrong thing to do to a territory that is already relatively hypoxaemic.
So the same two numbers Chapter 13 used as a low-output alarm are here a ventilator instruction: at ETCO₂ below 14 mmHg with a pulse pressure below 15 mmHg, turn the minute ventilation down and reconsider PEEP.
The Red Book adds the counterweight so that this is not read as licence: hypercarbia above 45 mmHg promotes pulmonary vasoconstriction, raising pulmonary vascular resistance and right ventricular afterload — and the right ventricle is what delivers transpulmonary flow in the first place.
Evidence — hyperoxia and hypocarbia are the default state on VA ECMO, not the exception
Two independent sources in the library make the same observation. The Red Book records that tissue hyperoxia (PaO₂ 100–300 mmHg) and hypocarbia (PaCO₂ below 30 mmHg) "are common on VA ECMO and may have deleterious side effects, including poor neurologic outcomes." Szuldrzynski's 2024 review states that hyperoxaemia above 300 mmHg and hypocarbia below 30 mmHg are common in VA support and "carry even greater burden given the risk of differential hypoxemia."
Certainty: moderate that both are common; low that they cause the neurological harm attributed to them — the association is observational and confounded by everything that puts a patient on VA ECMO.
The relevant randomised evidence is BLENDER (Chapter 14), which compared oxygen saturation targets of 92–96% against 97–100% in 300 VA ECMO patients and found no difference in any outcome. Note what that does and does not settle: it tested the systemic saturation target, not the regional problem this chapter is about, and it did not enrol patients on the basis of differential hypoxaemia.
Taha reports that ELSO's VA guidance suggests avoiding "excessive hypo- and hyperoxia" and monitoring right radial PaO₂, and that ECPR guidance recommends avoiding hyperoxia by blending fresh gas with air, targeting an arterial saturation of 92–97%.
16.7 Management — four levers, and only one of them is free
Every published account of this syndrome offers the same four levers. What the accounts disagree about is the order, and the disagreement is not cosmetic (§16.8).
Lever | What it does to the mixing point | What it costs |
1. Improve the native lung
FiO₂, PEEP, recruitment, bronchoscopy for secretions or plugging, diuresis, treating the infection, prone positioning, inhaled nitric oxide | Nothing. It leaves the mixing point exactly where it is and makes its position irrelevant, by oxygenating the blood the ventricle ejects | Nothing haemodynamic. This is the only lever with no trade-off — which is why every source lists it first |
2. Reduce native ejection
Raise circuit flow; reduce inotropes; reduce micro-axial pump flow; beta-blockade | Pushes the mixing point proximally, back towards the aortic valve, so more of the arch is circuit-perfused | Everything in Chapter 15. More afterload, less ejection, LV distension, pulmonary oedema, stasis and thrombus. You are deliberately suppressing the recovery you were waiting for |
3. Change where blood is drained from
Move or add venous drainage to the SVC or high right atrium | Removes deoxygenated upper-body venous return before it reaches the lungs, so what the ventricle does eject is better oxygenated | Another large-bore venous access; a repositioning procedure. No haemodynamic penalty — this is the quiet, underused lever (§16.9) |
4. Change where blood is delivered to
V-AV; subclavian or axillary return; central cannulation; conversion to VV | Abolishes the problem rather than displacing it: oxygenated blood now reaches the arch either through the lungs or antegrade from above | A procedure, a more complex circuit, flow-splitting problems, and — for subclavian and central — a substantially bigger operation (Chapters 12, 78) |
Clinical pearl — treat the lung first because it is the only move that is free
The ECPR text puts lever 1 first and in operational detail: "take steps to optimize pulmonary function by titrating mechanical ventilation to recruit the lungs and considering diuresis, antibiotics, and/or therapeutic bronchoscopy to treat any underlying lung problem to improve the oxygen saturation of blood ejected from the LV."
The largest clinical series supports the priority empirically. Among 44 patients with Harlequin syndrome, only 11 (25%) underwent ECMO revision; the other 33 (75%) were managed without it, and the authors conclude that "physiological interventions can effectively resolve Harlequin syndrome and should be considered as a key strategy before considering ECMO revision."
A published case makes the point concretely: bilateral lower-lobe atelectasis with thick purulent secretions, treated with bronchoscopy and lavage, PEEP of 10–12 cmH₂O and a negative fluid balance — oxygenation normal in all extremities by day 5, ECMO out on day 6.
Before you touch the flow, ask what is wrong with the lung and whether a bronchoscope would fix it.
Evidence — the largest series, and what it can and cannot tell us
Robinson J, Ejimogu J, Naselsky W, et al. Management of Harlequin syndrome in veno-arterial extracorporeal membrane oxygenation. Indian Journal of Thoracic and Cardiovascular Surgery. 2026;42(8):935–942. DOI 10.1007/s12055-026-02185-w.
Retrospective single-institution review, March 2016 to May 2021. 44 patients developed Harlequin syndrome. 11 (25%) underwent ECMO revision; 33 (75%) did not. Median time to resolution for the whole cohort 4.5 hours (IQR 1.4–7.80). Survival to decannulation 8/11 (73%) in the revision group versus 23/33 (70%) in the non-revision group; hospital length of stay similar. The revision group was substantially sicker at baseline: SAVE score −15 (IQR −17 to −11) versus −6.0 (IQR −10 to −1).
Certainty: low. Retrospective, single centre, no control group, and the two groups are not comparable — the SAVE difference tells you that revision was chosen for the sicker patients, which makes the near-identical survival uninterpretable in either direction. The denominator of VA ECMO patients was not stated, so this series yields no incidence figure. Full text retrieved.
What it does establish: that most episodes resolve without reconfiguration, and quickly. What it does not: whether any particular intervention caused that resolution, or whether earlier revision would have helped the patients who deteriorated.
This is also the second appearance of the SAVE score in this book, after BLENDER's subgroup analysis (Chapter 14). It remains uncovered — see the open items.
16.8 Controversy 1 — Should you raise the flow?
Controversy 1 — Is increasing ECMO flow the right response to differential hypoxaemia?
The question. Raising circuit flow drains more of the venous return, reduces transpulmonary flow and native ejection, and pushes the mixing point back towards the aortic valve. It is the classical second move after optimising the lung. But Chapter 15 established that raising flow is precisely how you distend a left ventricle — and this patient, by definition, has a ventricle that is ejecting against a pressurised aorta. The two chapters give opposite instructions for the same dial.
Position A — raise the flow. This is the textbook consensus. The Red Book: if right-arm saturation falls, "increasing peripheral VA flow (if possible) and thus decreasing flow through the pulmonary circulation may be sufficient to allow for adequate coronary and cerebral oxygenation." The ISCCM manual lists "increasing the pump flow" first among its three treatments. The ECPR text prescribes "maximizing ECMO flow to minimize native blood flow through the heart and lungs."
It also has the only controlled evidence in this chapter behind it. Rozencwajg S, Heinsar S, Wildi K, et al. Scientific Reports 2023;13:4002, DOI 10.1038/s41598-023-30226-6. Six sheep with ethanol-induced cardiogenic shock (ejection fraction below 30%, systolic pressure below 90 mmHg, lactate above 4 mmol/L) plus induced respiratory failure, on femoral VA ECMO, randomised to low flow 2.5 L/min or high flow 4.5 L/min for five hours. At high flow: brain tissue oxygen tension +215% from baseline versus −58% (p=0.043); cerebral NIRS 67 ± 5% versus 49 ± 4% (p=0.003); cerebral blood flow 970 ± 400 versus 205 ± 76 arbitrary perfusion units (p=0.042); and the primary outcome, global histological brain injury score, significantly worse in the low-flow group (p=0.0003), with neuronal shrinkage, congestion and perivascular oedema all p<0.0001. Cerebral microdialysis values in the low-flow group "all reached predefined pathological threshold." The authors conclude that differential hypoxaemia "can lead to cerebral damage after only a few hours" and that increasing flow "was an effective strategy to reduce such damages."
Position B — do not, or do it last. The Alfred Hospital ECMO guideline, updated 20 August 2026, orders the ladder differently. Step 1 is tolerate — recognise that the finding "represents a sign of cardiac recovery" and avoid intervention if the patient is improving. Step 2 is the lung. Increasing VA ECMO flow and MAP while reducing inotropy appears fifth, and is annotated as often unsuccessful, with thrombosis and pulmonary oedema risks. Reconfiguring the return cannula to the subclavian artery is described as "generally considered too aggressive for isolated differential hypoxia."
Taha states the mechanism of the objection: temporarily increasing flow "can notably be counterproductive, as this increases afterload against which the recovering left ventricle must work." And Chapter 15 supplies the arithmetic of what that costs.
What the evidence actually shows. Position A's experimental support is genuine and should not be waved away: it is randomised, it has a hard histological endpoint, and the effect sizes are large. But its limits are equally real, and its own authors state the decisive one: the study "did not assess mortality or long-term outcome," so the cerebral benefit "may not outweigh LV damage from increased afterload." Six animals, three per group, five hours, healthy young sheep. Certainty: low — an internally valid experiment in a model, not evidence about patients.
Position B has no controlled evidence at all. It has a mechanism, a 2026 unit guideline, and the largest clinical series showing that three-quarters of episodes resolve without reconfiguration.
The resolution this book proposes. The two positions are not really disagreeing about physiology. They are disagreeing about which organ is currently closer to being damaged, and that is a patient-level question, not a doctrinal one.
— Raising flow buys the brain oxygen by taking ejection away from the ventricle. In a patient with a shut or barely opening aortic valve, that price is unpayable — you convert a hypoxaemia problem into a distension problem, and Chapter 15's cascade is not slower or more forgiving than this one.
— Lowering flow buys the ventricle ejection at the price of pushing native, poorly oxygenated blood further up the aorta. In a patient with a right radial saturation in the seventies, that price is unpayable too.
The single most important observation in this chapter is that both problems have the same cause — a ventricle ejecting poorly oxygenated blood — and only two interventions treat the cause rather than trading one organ against the other: fixing the lung, and changing the circuit. Flow adjustment is a temporising measure in both directions. It should be recognised as such, used deliberately and briefly, and accompanied by work on a definitive answer.
Certainty: moderate that raising flow improves cerebral oxygenation; very low that it improves any patient outcome; moderate that it worsens ventricular loading.
Where practice actually sits. Most units raise flow as a bridge while preparing something better, exactly as one published case describes: beta-blockade plus increased flow to buy "acceptable arterial oxygen saturations whilst we converted the circuit to VAV-ECMO."
What would resolve it. A trial randomising flow strategy in patients with established differential hypoxaemia, with cerebral and ventricular endpoints measured together. Nothing of the kind is registered. The nearest existing instrument is the combination Chapter 15 already recommends — echocardiographic aortic valve opening alongside right radial saturation — used as a two-sided constraint rather than a single target.
What to take to the bedside. Do not turn the flow up without looking at the aortic valve, and do not turn it down without looking at the right radial gas. If the valve is opening and the right side is hypoxaemic, you have room to raise the flow. If the valve is barely opening and the right side is hypoxaemic, you have run out of room on this dial and the answer is the lung or the circuit.
Pitfall — a micro-axial pump makes this syndrome worse, and Chapter 15 may have put one there
A 2025 comprehensive review of Harlequin syndrome in VA ECMO and ECPELLA makes the interaction explicit: adding a micro-axial pump, "although beneficial for ventricular unloading and hemodynamic support, may further exacerbate this competition and precipitate" differential hypoxaemia.
The mechanism is unavoidable. The device actively propels blood from the left ventricle into the ascending aorta — and if the lungs are failing, that blood is poorly oxygenated. A device chosen to protect the ventricle pushes the mixing point distally and puts the brain at risk.
The ECPR text gives the corresponding instruction for a patient who has both: limit "inotrope doses and antegrade LV vent flow (i.e. Impella flow) to the minimum necessary to prevent LV distension." That sentence is a compromise, and it should be read as one — the minimum necessary, not the maximum tolerated.
Certainty: low — mechanistic reasoning supported by case reports and a narrative review; no comparative data. The practical consequence is concrete: if you unload a ventricle with a micro-axial pump in a patient with bad lungs, put a right radial gas and cerebral oximetry on the plan for the same shift.
16.9 Controversy 2 — Prevent it at cannulation, or detect it and react?
Controversy 2 — Should the venous drainage site be chosen to prevent differential hypoxaemia?
The question. The syndrome exists because deoxygenated blood from the upper body returns through the superior vena cava, is not captured by a femoral drainage cannula sitting in the inferior cava, crosses the lungs unoxygenated and is ejected back into the arch. Draining the SVC instead removes that blood before it ever reaches the failing lung. So should cannulation be planned around this, or should the syndrome be detected and treated when it appears?
The case for prevention. A 2020 Critical Care correspondence argues it directly. Reporting a simulation study combined with a case report, the authors state that changing the venous cannulation site from the inferior to the superior caval vein increased right-arm arterial saturation from below 60% to above 80%, and that "arterial saturations are highly dependent on the venous cannulation site during VA-ECMO in patients with severe respiratory failure." Their conclusion is a rhetorical question and its answer: "should we pay more attention to detecting the watershed or should we focus on applying better cannula configurations... We believe that we should apply better cannula configurations." They note it can be achieved simply — a jugular cannula with a high atrial tip, or a long femoral venous cannula without side holes reaching the upper right atrium.
The Red Book records the same idea more cautiously: "strictly draining deoxygenated venous blood from the superior vena cava has been proposed to mitigate differential hypoxemia."
The case against making it routine. Differential hypoxaemia affects a minority of femoral VA ECMO patients and needs the coincidence of three conditions (§16.1). Most VA cannulation is done urgently or during resuscitation, where the femoral route is chosen because it is fast and does not require the operator to leave the groin (Chapters 12, 21). Optimising every cannulation against a complication that most patients will not develop imposes a cost — a second site, a longer procedure, a different cannula — on everybody.
What the evidence actually shows. The supporting data are a simulation plus a single case, reported second-hand in a correspondence article, and the primary paper was not retrieved in this session. Certainty: very low. But the physiology is not in doubt: SVC drainage removes the exact blood that causes the syndrome, and the ISCCM manual independently notes that adding a venous cannula is indicated both for inadequate drainage and for differential hypoxia.
The resolution this book proposes. This is not one question but two, and they have different answers.
— For an unselected VA ECMO cannulation, no. Speed and simplicity win, and the syndrome is uncommon.
— For a patient cannulated with known severe lung disease — ARDS, aspiration, pulmonary haemorrhage, dense consolidation — the calculus reverses, because all three preconditions for the syndrome are already met on day zero and the only missing ingredient is cardiac recovery, which is the thing you are hoping for. In that patient, the drainage plan is a decision, not a default, and a tip position high in the right atrium costs almost nothing to choose deliberately.
What to take to the bedside. Ask at cannulation: if this heart recovers next week, will the blood it ejects be worth ejecting? If the answer is no, plan the drainage accordingly and site the right radial line before you start.
16.10 Changing the circuit — which third cannula, and where
When the lung cannot be fixed fast enough, the answer is a configuration change. The commonest error here is to reach for the familiar option rather than the one that matches the mechanism.
Physiology — the third cannula does one of two entirely different jobs
The Red Book's paediatric cardiac chapter states the distinction more clearly than most adult sources: an additional venous cannula may be placed "either as an additional drainage cannula to increase total circuit flow or as a return cannula to oxygenate blood as it traverses the pulmonary circulation."
Those are different operations for different problems.
— Extra drainage (VV-A) — the added cannula takes blood out. It raises achievable circuit flow and removes deoxygenated upper-body return. Use it when the limit is drainage — a small cannula, a large patient, or, as the 2021 Perfusion case showed, right ventricular dysfunction.
— Extra return (V-AV) — the added cannula puts oxygenated blood in, into the right atrium, so that the blood crossing the lungs is already oxygenated before the ventricle ejects it. Use it when the limit is the lung, which is the usual case.
Getting this backwards is a real error. A patient whose problem is a failing right ventricle does not need oxygenated blood delivered into a right heart that cannot move it.
Option | How it fixes the syndrome | Cost and caveats |
V-AV (veno-arterial-venous) — arterial return split, with a limb to the right internal jugular vein | Oxygenates blood before it crosses the lungs, so native ejection is no longer hypoxaemic. The commonest definitive answer, and the one most case reports use | Flow splitting. Both the Red Book and the ISCCM manual warn that flow preferentially takes the low-resistance venous limb, "leading to systemic arterial hypoperfusion." Needs a Y-connector, adjustable clamps and flow sensors on both limbs. The Alfred guideline advises maintaining at least 1 L/min in each limb (Chapter 78) |
VV-A (veno-veno-arterial) — additional drainage cannula | Raises flow and strips deoxygenated upper-body venous return | Does not oxygenate transpulmonary blood. The right answer when the problem is drainage or the right ventricle, the wrong one when it is the lung |
Move venous drainage to the SVC or high right atrium | Same mechanism as VV-A without a third cannula (§16.9) | A repositioning procedure under live imaging. Evidence is a simulation and a case |
Subclavian or axillary arterial return | Antegrade delivery into the arch — the mixing point disappears entirely | Surgical, usually a side graft. Introduces upper-limb hyperperfusion, compartment syndrome risk and nerve injury (Chapters 12, 17). The Alfred guideline considers it "too aggressive for isolated differential hypoxia" |
Central cannulation | Abolishes the watershed and allows direct venting (Chapter 15). The ISCCM manual: central cannulation gives "excellent antegrade flow delivery into the arterial system, thus offsetting the watershed phenomenon" | Sternotomy, bleeding, infection, immobility. Reserved for patients who need it for other reasons too |
Conversion to VV | The definitive answer if the heart has genuinely recovered — which the syndrome itself is evidence for | Requires that cardiac function is sufficient to stand alone. Assess formally (Chapter 18) rather than assuming |
Return cannula into the main pulmonary artery | A 2025 case report describes a 17 Fr single-stage cannula placed via the left subclavian vein into the main pulmonary artery and attached to the arterial limb, which resolved the syndrome and "provides both oxygenation and hemodynamic support for the right ventricle" | Certainty: very low — a single case. Recorded because it is mechanistically elegant, not because it is established |
Clinical pearl — buying time while the circuit is being rebuilt
Reconfiguration takes preparation. A 2023 case series in European Heart Journal: Acute Cardiovascular Care describes the bridging problem honestly — conversion to V-AV "takes some preparation and can be time-consuming," and their intention was "to promptly improve oxygen saturation while simultaneously working on a more definitive solution."
Their bridge was intravenous beta-blockade to reduce native left ventricular output, together with increased circuit flow. It worked: acceptable saturations while the V-AV circuit was assembled. Another published case lists the same option — "beta blockers to reduce native cardiac output till improvement in pulmonary function."
Pause on what that means. You are deliberately suppressing the cardiac recovery you have spent a week working for, in order to stop the recovering heart poisoning the brain. It is the sharpest illustration in this book that VA ECMO support and cardiac recovery are not the same goal.
Certainty: very low — case reports only, with no comparative data and no guidance on agent or dose. Use it as a short bridge to a definitive plan, not as a management strategy.
16.11 The errors that recur
Error | Correction |
Sampling arterial gases from a femoral or left-sided line | That is circuit blood. The right radial samples what the brain and coronaries receive |
Treating a normal right radial gas as excluding the syndrome | The mixing point can sit proximal to the brachiocephalic trunk. Then only the coronaries are affected, and the ECG is the monitor |
Reading a new ST elevation on femoral VA ECMO as coronary occlusion | It may be isolated cardiac Harlequin. The treatment is a venous return limb, not a stent |
Concluding the saturation is unmeasurable because the SpO₂ trace is poor | Low pulsatility breaks pulse oximetry precisely when this syndrome is most likely. Use NIRS and a right radial gas |
Reconfiguring the circuit before checking the oxygenator | One paired post-oxygenator and right radial sample separates distribution failure from membrane failure |
Raising the flow without looking at the aortic valve | You are trading the ventricle for the brain. Chapter 15 describes what you are buying with |
Lowering the flow to protect the ventricle without checking the right radial gas | The same trade in the other direction. Both dials need both numbers |
Treating flow adjustment as the definitive answer | It displaces the mixing point; it does not oxygenate the blood. Only the lung and the circuit do that |
Adding a venous return limb when the problem is a failing right ventricle | Oxygenated blood delivered into a right heart that cannot move it achieves nothing. That patient needs drainage — VV-A |
Building a V-AV circuit without flow sensors and clamps on both limbs | Flow steals to the low-resistance venous limb and the systemic circulation is underperfused |
Adding a micro-axial pump for LV unloading and not rechecking upper-body oxygenation | The device propels poorly oxygenated blood into the ascending aorta. It can create this syndrome |
Leaving the ventilator on a normal minute ventilation when transpulmonary flow is minimal | The little blood crossing the lung becomes severely alkalotic, and it is the blood going to the brain |
Reading a filling defect at the arch on contrast CT as thrombus or dissection | Repeat the acquisition at reduced ECMO flow. It is usually the mixing cloud |
16.12 Key points
- Differential hypoxaemia is a sign of cardiac recovery. It needs a ventricle that ejects, lungs that do not work, and retrograde femoral return. Remove any one and it cannot happen.
- The territory above the watershed is the coronary and cerebral circulation — the two organs the whole run exists to protect.
- It is fast. The only controlled experimental study found histological cerebral injury after a few hours. Treat a new right-sided desaturation as an emergency.
- On femoral VA ECMO there is no such thing as "the" arterial blood gas. Right radial is the reference site; lower-body and circuit samples describe the machine.
- Three simultaneous gases — right radial, left radial, circuit — localise the mixing cloud and tell you how much of the arch is native.
- A normal right radial gas does not exclude the syndrome. With the mixing point in the ascending aorta, only the coronaries are affected, and the presentation is ST elevation or an electrical storm.
- Pulse oximetry is least reliable exactly when this syndrome is most likely, because low pulsatility is a consequence of the high flow that also drives the mixing point distally. NIRS does not need a pulse.
- One paired sample — post-oxygenator and right radial — separates a distribution problem from a membrane problem.
- The differential includes differential venous drainage, right ventricular failure, cannula malposition, intracardiac shunt and reverse harlequin. Not every hypoxaemic upper body is a watershed.
- The same anatomy produces a carbon dioxide gradient. Upper-body PaCO₂ is set by the ventilator, lower-body PaCO₂ by the sweep gas, and the respiratory centre sits in the upper-body territory — so the patient's own drive is regulated by only one of the two compartments.
- At transpulmonary flow below about 1 L/min — pulse pressure under 15 mmHg with end-tidal CO₂ under 14 mmHg — turn the minute ventilation down, or the small volume of blood reaching the brain becomes severely alkalotic.
- Fixing the lung is the only lever with no haemodynamic cost. Three-quarters of episodes in the largest series resolved without reconfiguration, at a median of 4.5 hours.
- Flow adjustment is a temporising measure in both directions. Raising it protects the brain at the ventricle's expense; lowering it does the reverse. Neither oxygenates any blood.
- Do not turn the flow up without looking at the aortic valve, and do not turn it down without looking at the right radial gas.
- The third cannula does one of two different jobs. Extra return (V-AV) oxygenates transpulmonary blood — the usual answer. Extra drainage (VV-A) raises flow and strips upper-body venous return — the answer when the right ventricle or the drainage is the problem.
- In V-AV, flow steals to the low-resistance venous limb. Clamps, flow sensors, and a floor on each limb are not optional.
- A micro-axial pump placed to unload the ventricle can precipitate this syndrome, because it propels poorly oxygenated blood into the ascending aorta.
- Ask at cannulation whether this patient's lungs would make recovery dangerous. If so, the venous drainage plan and the right radial line are decisions, not defaults.
[VERIFICATION REQUIRED] — open items in this chapter
- Retrieval was curtailed. The session's full-text fetch quota was exhausted partway through the external search. Several sources below were therefore read only as published abstracts or as reported in a retrieved secondary source, and are marked as such. The affected items should be checked before any of their numbers are reused.
- The 8.8% incidence is quoted from a 2024 paper that itself says the incidence "has not been clearly stated", and from a 2025 case report citing the same upstream reference. That upstream reference was not retrieved. No independent incidence estimate was found, and §16.2 argues the figure should not be trusted as a rate of occurrence.
- Rozencwajg 2023 (Scientific Reports 13:4002) was retrieved in full. Its limitations are the authors' own: three animals per group, five hours, healthy young female sheep, no mortality or long-term outcome, and an explicit statement that the cerebral benefit may not outweigh ventricular harm.
- Robinson 2026 (Indian J Thorac Cardiovasc Surg 42(8):935–942) was retrieved in full, but the article as retrieved does not state the denominator of VA ECMO patients, the diagnostic criteria used, or which interventions were applied to how many patients. No incidence and no intervention-level conclusion can be drawn from it.
- The radial NIRS proposal and the SVC-drainage simulation were read as abstracts or through a retrieved secondary source. The simulation's headline figures — right-arm saturation rising from below 60% to above 80% — are as reported in a correspondence article; the primary study was not retrieved, and its design, sample and modelling assumptions are unknown.
- The isolated cardiac Harlequin case, the differential venous drainage series, the right-ventricular case, the pulmonary artery return cannula, the CT mixing-cloud case, the beta-blockade bridge and the three-site sampling technique are all single case reports or small case series, read as abstracts. They are included because each describes a failure mode or manoeuvre absent from the textbooks, not because any is established practice. Full author lists, volumes and pages are not asserted.
- The Alfred Hospital ECMO guideline is local institutional guidance updated 20 August 2026, not a society document. It is cited as evidence that a credible unit orders the ladder differently, not as a recommendation.
- ELSO's adult VA guidance on right radial PaO₂ monitoring, and the ECPR guidance target of SaO₂ 92–97%, are as reported by Taha; the guideline documents themselves have not been retrieved in this project (see the standing item in the book status file).
- The Red Book's differential carbon dioxide account is cross-referenced by that book to material on the ELSO website which was not retrieved. No clinical outcome data on differential carbon dioxide were found.
- The transpulmonary flow below 1 L/min rule reuses the ETCO₂ and pulse-pressure thresholds whose single underlying reference remains unretrieved (Chapter 13's standing item).
- Post-oxygenator PaO₂ 150–300 mmHg, PEEP 10–12 cmH₂O, respiratory rate 6–12/min, the 1 L/min per limb V-AV floor and the 85% circuit inlet saturation caution are all expert or local conventions without cited supporting outcome data.
- The resolution offered in §16.8 — that flow adjustment is a temporising trade between two organs and that only the lung and the circuit treat the cause — is this book's reasoning, not a published position. The two positions it reconciles are cited; the reconciliation is not.
- The §16.9 distinction between routine cannulation and cannulation in known severe lung disease is likewise this book's judgement, offered as a question to ask rather than a rule.
- The SAVE score appears here for the second time in this book and is still not covered anywhere. It remains an open gap.
Cross-references
- Chapter 2 — Cardiopulmonary Physiology for ECMO: the mixing equation and oxygen content, which this chapter applies regionally
- Chapter 9 — Persistent Hypoxaemia on VV ECMO: the same presenting problem in the configuration where a watershed cannot exist. Worth reading alongside for the contrast
- Chapter 12 — VA ECMO Cannulation: the arterial site as the decision that creates or prevents this syndrome, and the right radial line
- Chapter 13 — VA ECMO Haemodynamics: §13.2 the moving mixing point; §13.4 the reverse harlequin effect and the pink patient with an ischaemic ECG; §13.6.1 the ETCO₂ and pulse-pressure rule reused here
- Chapter 14 — Initial VA ECMO Management: BLENDER and oxygen targets; the first-hour instruction to sample right-sided
- Chapter 15 — LV Distension and LV Unloading: the opposite consequence of the same physics, and the other end of the flow dial
- Chapter 17 — Limb Ischaemia and Vascular Complications: upper-limb hyperperfusion after subclavian or axillary return
- Chapter 18 — VA ECMO Weaning and Decannulation: whether the recovery this syndrome signals is enough to stand alone
- Chapter 26 — Echocardiography During ECMO: aortic valve opening and right ventricular assessment
- Chapter 27 — Blood Gas and Oxygenation Monitoring: sampling protocols
- Chapter 28 — Cerebral Monitoring: NIRS, transcranial Doppler and their confounders
- Chapter 32 — Oxygenator Failure: the first differential
- Chapter 53 — Mechanical Ventilation During ECMO: the ventilator strategy this chapter leans on
- Chapter 78 — Hybrid ECMO Configurations: building and running V-AV and VV-A circuits
- Chapter 80 — ECMO + Impella: the interaction in §16.8
References
Experimental
- Rozencwajg S, Heinsar S, Wildi K, et al. Effect of flow change on brain injury during an experimental model of differential hypoxaemia in cardiogenic shock supported by extracorporeal membrane oxygenation. Scientific Reports. 2023;13:4002. DOI 10.1038/s41598-023-30226-6. Retrieved in full.
Clinical series
- Robinson J, Ejimogu J, Naselsky W, et al. Management of Harlequin syndrome in veno-arterial extracorporeal membrane oxygenation. Indian Journal of Thoracic and Cardiovascular Surgery. 2026;42(8):935–942. DOI 10.1007/s12055-026-02185-w. Retrieved in full; denominator and intervention detail not reported.
Monitoring, mechanism and technique — all read as abstracts or through a retrieved secondary source; author lists, volumes and pages not asserted
- Yu Y, et al. To identify Harlequin syndrome in patients with venoarterial extracorporeal membrane oxygenation using radial near-infrared spectroscopy. Critical Care. 2024. DOI 10.1186/s13054-023-04793-z.
- Torre DE, et al. Harlequin Syndrome in Venoarterial ECMO and ECPELLA: When ECMO and Native or Impella Circulations Collide — A Comprehensive Review. Reviews in Cardiovascular Medicine. 2025. DOI 10.31083/rcm39992.
- Giunta M, et al. Management of Harlequin Syndrome Under ECPELLA Support: A Report of Two Cases and a Proposed Approach. Annals of Cardiac Anaesthesia. 2023. DOI 10.4103/aca.aca_176_21. Three-site sampling to localise the mixing cloud.
- Honore P, et al. Risk of harlequin syndrome during bi-femoral peripheral VA-ECMO: should we pay more attention to the watershed or try to change the venous cannulation site? Critical Care. 2020. DOI 10.1186/s13054-020-03168-y. Reports the SVC-drainage simulation and case; the primary study was not retrieved.
- Heinrich MA, et al. Case report of a cardiac Harlequin syndrome — electrical storm during venoarterial extracorporeal membrane oxygenation. European Heart Journal: Case Reports. 2025. DOI 10.1093/ehjcr/ytaf059.
- Heymer J, et al. Differential Venous Drainage Mimicking Differential Oxygenation in Patients on Extracorporeal Life Support. ASAIO Journal. 2025. DOI 10.1097/MAT.0000000000002420.
- Wilson J, et al. Managing Harlequin Syndrome in VA-ECMO — do not forget the right ventricle. Perfusion. 2021. DOI 10.1177/02676591211020895.
- De Ridder S, et al. Answer: Overthrowing the Harlequin. European Heart Journal: Acute Cardiovascular Care. 2023. DOI 10.1093/ehjacc/zuad037. The beta-blockade bridge.
- Antonogiannakis A, et al. Successful management of harlequin syndrome due to pulmonary hemorrhage and atelectasis with VAV-ECMO. Perfusion. 2023. DOI 10.1177/02676591231181847.
Guidance
- Alfred Hospital ECMO Guideline — Differential hypoxia. Updated 20 August 2026. Local institutional guidance, not a society document.
- Szuldrzynski K, et al. Mechanical ventilation during extracorporeal membrane oxygenation support: new trends and perspectives. Perfusion. 2024. Source for the post-oxygenator PaO₂ target and for the reported content of the 2021 ELSO Interim VA guideline. Held in the project library and read in full.
Textbooks
- ELSO Red Book, 6th edition, Chapter 2 (nomenclature; differential hypoxia and differential carbon dioxide), Chapter 4 (hybrid configurations and the low-resistance-limb steal), Chapter 18 (paediatric cardiac — the two roles of a third cannula), Chapter 28 (detection by right-arm oximetry and forehead NIRS; V-VA, subclavian and central options; SVC drainage; ventilation on VA ECMO and the transpulmonary flow rule). [VERIFICATION REQUIRED] — page numbers not confirmed.
- ISCCM Manual of RRT and ECMO in ICU, Chapter 25 (watershed position; VAV flow splitting), Chapter 30 (triple cannulation; central cannulation offsetting the watershed), Chapter 41 (the three-item treatment list). [VERIFICATION REQUIRED].
- Taha AR, Caridi-Scheible M, Leiendecker E, et al. ECMO: A Practical Guide to Management, Chapter 6 (oxygenation and hypoxia on VA ECMO; the counterproductive-flow warning; the monitoring table and NIRS confounders) and Chapter 7 (configuration complications). [VERIFICATION REQUIRED].
- Shinar Z, Badulak J, eds. ECPR and Resuscitative ECMO, Chapter 7 (Figure 9 and the full management sequence, including limiting inotropes and micro-axial pump flow). [VERIFICATION REQUIRED].
Chapter status
Drafted and audited 7 September 2026. Ten-pass quality control completed: clinical, physiology, evidence, citation, numerical, safety, contradiction, redundancy, bedside utility and literature-currency passes.
The contradiction audit was the substantial one, and unusually it was internal. Chapter 15 §15.5 makes lowering ECMO flow the first medical step for a distending ventricle; every classical source on differential hypoxaemia makes raising it the first step after treating the lung. §16.8 sets both out with their evidence — including a randomised ovine experiment supporting the flow increase and a 2026 unit guideline placing it fifth — and resolves them by noting that the two complications share one cause and that flow adjustment only trades one organ against the other. That resolution is labelled as this book's reasoning. A dated addendum has been appended to Chapter 15 recording the conflict from the other side.
The redundancy audit removed the mixing-point physics, which belongs to Chapter 13 §13.2, and the hybrid-circuit build detail, which belongs to Chapter 78.
The literature-currency pass is the weakest of the ten in this chapter: the session's full-text retrieval quota was exhausted partway through, and the verification callout records exactly which sources were read only as abstracts.